visa-doc-translate

Pass

Audited by Gen Agent Trust Hub on Apr 6, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the 'sips' utility for image conversion and 'brew' for managing system-level OCR dependencies.
  • [EXTERNAL_DOWNLOADS]: It installs several standard third-party libraries including 'easyocr', 'pytesseract', and 'reportlab' via pip and brew.
  • [DYNAMIC_EXECUTION]: The skill generates a custom Python script to create a formatted PDF from the original image and translated text, which is then executed locally.
  • [INDIRECT_PROMPT_INJECTION]: By performing OCR on user-provided images, the skill ingests untrusted data into the agent context. This creates a potential surface for adversarial instructions embedded in documents. Evidence: 1. Ingestion points: user-provided image files processed in SKILL.md. 2. Boundary markers: None. 3. Capability inventory: sips, pip, brew, and Python script execution. 4. Sanitization: None mentioned.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 6, 2026, 04:01 AM
Security Audit — agent-trust-hub — visa-doc-translate